Educational research tool · Nothing here is advice

My backtest said +1650%. I picked the tickers with hindsight.

Every retail backtester would have shown me that first number and none of the corrections. So I built the one that argues back — it runs a full honesty audit on every result, and there is no way to turn it off.

The number came from forty-four tickers I picked in 2026 and tested on prices from February 2023 to May 2026. I was buying winners I already knew had won. A stricter strategy on the full S&P 500 made 9.8% a year from 2016 to 2026, behind SPY's 16%. A dip-buying strategy beat SPY on Sharpe before costs, then fell behind it at five basis points a day. Three numbers, one afternoon. The audit that caught them is now the product.

What my backtest said

+1650%

136% a year over 40 months, on 44 tickers picked with hindsight. Max drawdown 45%, against SPY's 8%.

A stricter strategy, full S&P 500

+163%

Different rules: momentum plus low volatility, top 25 names, 2016 to 2026. SPY returned +362% over the same years.

After trading costs

Lost

A dip-buyer beat SPY on Sharpe before costs, 1.05 to 0.83. At five basis points a day it was still up 103% (Sharpe 0.41), behind SPY's 344% (0.83).

Watch it argue back

One sentence goes in. An argument comes back.

The strategy “Each month, buy the 5 strongest stocks in my 44-ticker watchlist by 6-month return. Equal weight.” is read back as a spec — top 5, 126-day return, monthly rebalance, equal weight, vs spy — and audited: five checks pass, one warns that the 44-ticker universe is hand-picked. Same rules, full S&P 500: +960.2% vs your +1816.5%. Hand-picking alone: 1.81× in final wealth. Verdict: amber — Beats the benchmark — but your universe is hand-picked. Strategy versus SPY buy-and-hold: Total return +1816.5% vs +41.9%, CAGR +371.3% vs +20.2%, Sharpe 2.07 vs 0.92, Max drawdown −60.1% vs −18.8%.

04 · Verdict

Wounded — amber verdict

Beats the benchmark — but your universe is hand-picked.

Total return: +1816.5%; SPY +41.9%
CAGR: +371.3%; SPY +20.2%
Sharpe: 2.07; SPY 0.92
Max drawdown: −60.1%; SPY −18.8%

Strategy ×19.2SPY ×1.42log scale · 2024-07-01 → 2026-05-29

Benchmark shown on every result. No way to turn the audit off.

A replay of run rep_2c6332f417, the same run as the export further down this page. It is a separate test on the same 44 tickers, ranked by 6-month return from mid-2024 to May 2026, so its numbers differ from the backtest above. Every number is from the real report.

Every run ends here

The verdict comes first. The flattering number comes second.

Plain English goes in; a precise, editable spec comes back for you to approve. Nothing runs until you confirm exactly what will be tested — and what comes out is this.

An exported Truth Report: an amber verdict reading “Beats the benchmark — but your universe is hand-picked”, above the strategy's return and Sharpe shown next to the SPY benchmark and the full cost sweep table. — A real export from the running application, not a mock-up. The benchmark row and the disclaimer are baked into the image file itself, so a screenshot on social media carries them too.

Six ways it tries to kill your idea

The audit is not a feature. It is the only path to a result.

Selection bias, named out loud

Hand-pick a watchlist and your exact rules are re-run on the full S&P 500 for comparison. Hindsight gets reported, not rewarded.

Costs swept at 0, 5 and 20 bps

Scaled by your strategy's measured turnover. The report states the cost level at which the edge dies — often lower than you'd guess.

A verdict you didn't get to see

Thirty per cent of the window is held out. A large in-sample advantage that vanishes there is flagged as the overfit signature it is.

Lookahead made impossible

Signals at day t, returns from t+1, enforced inside the engine rather than by convention. Altering future prices cannot change past weights.

Regime coverage, stated plainly

If your window contains no sustained bear market, the report says so. A strategy tested only through a bull run has not been tested.

Reproducible to the byte

Every run freezes its price panel under a content hash and replays it exactly. Your journal cannot quietly drift underneath you.

Terms

Free forever, for the part that matters

The audit is never paywalled. Charging you to discover that your strategy is broken would invert the point of the product — so the honesty is free, and the paid tier is better data, which is where the real cost actually sits.

Free

$0 / forever

The complete honesty audit, unabridged.

✓ Every audit check — selection bias, cost sweep, regimes, holdout, sensitivity
✓ Benchmark on every result
✓ Plain-English to spec translation
✓ Shareable Truth Report exports
✓ Strategy journal and version history
— Current index membership only
— Monthly backtest allowance

Use it now — no account

Pro — planned

TBD / month

Survivorship-free data and forward tracking. Priced once the data licence is signed, not before.

✓ Point-in-time index membership, including companies that went to zero
✓ Delisted-stock coverage back through 2008
✓ Forward testing — freeze a survivor, track it live against its backtest
✓ Unlimited backtests and longer history
✓ Everything in Free

Join the waitlist

No pricing is committed, and nothing on this page is for sale. The paid tier ships only after a licensed data source and a legal review are in place.